feat(tcloud): imagesEdit — OpenAI /v1/images/edits over multipart#31
Open
drewstone wants to merge 1 commit into
Open
feat(tcloud): imagesEdit — OpenAI /v1/images/edits over multipart#31drewstone wants to merge 1 commit into
drewstone wants to merge 1 commit into
Conversation
Adds the sibling to `imageGenerate` so callers can drive the
edit / inpaint / multi-image composition workflow without
hand-rolling a custom transport. cli-bridge already speaks the
target route (`POST /v1/images/edits` proxied to either
`router.tangle.tools/v1/images/edits` when TANGLE_API_KEY is set
or OpenAI direct), so this is the last piece needed for the
"place this CAD design into the user's uploaded yard photo"
workflow in physim's `generate_concept_image`.
API:
client.imagesEdit({
prompt: "...",
image: blob | ArrayBuffer | {data: b64, mediaType, filename?},
mask?: <same shape>, // inpainting mask
model?: "gpt-image-2", // default
n?: 1,
size?: "1024x1024",
quality?: "high",
response_format?: "b64_json",
}) → ImageResponse
The `image` field accepts a single attachment or an array (for
gpt-image-2's multi-image fusion). Polymorphic input keeps both
browser (Blob / ArrayBuffer from `fetch().then(r => r.blob())`)
and Node-server callers (base64 strings from a DB / disk read)
ergonomic without forcing either side to do a conversion dance.
The transport mirrors the existing `transcribe()` path —
multipart/form-data through `proxiedFetch`, `Content-Type`
deleted so FormData sets the boundary, manual `checkLimits()` +
`_requestCount++` for credit accounting.
Bumps to v0.4.7. Re-exported from `index.ts` alongside the
existing image types.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds the sibling to
imageGenerateso callers can drive the OpenAI image-edit / inpaint / multi-image-fusion workflow without hand-rolling a custom multipart transport. Closes the last gap blocking physim's "place this CAD design into the user's uploaded yard photo" feature.API
```ts
import { TCloud, type ImageEditOptions } from '@tangle-network/tcloud'
await client.imagesEdit({
prompt: 'A chicken coop placed in this yard, snowy Vermont, golden hour',
image: yardPhotoBlob, // or ArrayBuffer, or {data: b64, mediaType, filename?}
mask: roiMaskPng, // optional inpainting mask
model: 'gpt-image-2', // default
n: 1,
size: '1024x1024',
quality: 'high',
response_format: 'b64_json',
})
```
`image` is polymorphic: single attachment OR an array (gpt-image-2 supports multi-image fusion — pass two reference frames + a prompt that composes them). Accepted shapes:
Transport
Mirrors the existing `transcribe()` pattern — multipart through `proxiedFetch`, `Content-Type` deleted so FormData sets its own boundary, manual `checkLimits()` + `_requestCount++` for credit accounting. cli-bridge already implements the target route (`/v1/images/edits`) and dispatches to either `router.tangle.tools` or OpenAI direct.
Why now
Used by physim's generate_concept_image once this lands — the tool currently routes through `imageGenerate` for prompt-only renders; the edit path lets it ingest a user-uploaded reference photo (yard, room, workshop) and place the proposed CAD design into it. The image stays consistent with the SCAD parameters by construction because the prompt is assembled from the same source-of-truth spec.
Version bump 0.4.6 → 0.4.7 (additive — existing surface unchanged).